Hey @Shubham Bajaj @Sahil I just tested another call ccccd7af-f15f-4ad6-91e4-c3475b686c5c It seems like I see in the logs "I can hear you perfectly" but as you can tell on the audio, we don't hear the bot at all. Do you know what happened? Oh also, when are you planning to have Claude 3.7? Thanks!

Hey there! 😊 It sounds like you're experiencing an audio issue with the bot during your call. Here are a few potential reasons for why you might not be hearing the bot, even though the logs indicate it's functioning: 1. **Service Interruptions**: There could be a service degradation with Deepgram, which is responsible for audio processing. It's a good idea to check their status page for any ongoing issues. 2. **Configuration Errors**: Sometimes, issues with microphone permissions or call settings can lead to audio not being transmitted properly. Make sure everything is configured correctly. 3. **Audio Feedback Issues**: If you're using speakerphone, it might cause echoing, which can confuse the AI and disrupt the conversation. 4. **Network Issues**: High latency or network restrictions can also affect audio quality. Testing in a different network environment might help. As for Claude 3.7, I couldn't find any specific release date information. Currently, there are no official announcements regarding its launch. It's best to keep an eye on official channels for updates. If you have any more questions or need further assistance, feel free to ask! Sources: - [Audio Issues with Voice AI Bots](https://discord.com/channels/1211482211119796234/1211483291191083018/1296852804631265392) - [Expected Release Date for Claude 3.7](https://discord.com/channels/1211482211119796234/1211483291191083018/1340022199687577680)

current_config

{"waitSeconds": 1.5, "smartEndpointingEnabled": true, "transcriptionEndpointingPlan": {"onPunctuationSeconds": 0.8}}

{"model": "nova-2-general", "language": "multi", "provider": "deepgram"}

next_config: with multi

{"waitSeconds": 0.8, "smartEndpointingEnabled": false}

{"model": "nova-2", "language": "multi", "provider": "deepgram", "endpointing": 300}

next_config: w/o multi

{"waitSeconds": 0.8, "smartEndpointingEnabled": true}

{"model": "nova-2", "language": "en", "provider": "deepgram", "endpointing": 300}

@Shubham Bajaj Sorry I'm just getting back on the horse. So are you saying that when I have multi enabled I should disable smartEndpointing?

Yes only when using non-english language .